A&A 625, A122 (2019)https://doi.org/10.1051/0004-6361/201935368c© ESO 2019

Astronomy&Astrophysics

CIELO-RGS: a catalog of soft X-ray ionized emission lines?

Junjie Mao1,2, Jelle S. Kaastra1,3, Matteo Guainazzi4, Rosario González-Riestra5, Maria Santos-Lleó5,Peter Kretschmar5, Victoria Grinberg6, Eleni Kalfountzou5, Aitor Ibarra5, Gabi Matzeu5,

Michael Parker5, and Pedro Rodríguez-Pascual5

1 SRON Netherlands Institute for Space Research, Sorbonnelaan 2, 3584 CA Utrecht, The Netherlandse-mail: j.mao@sron.nl

2 Department of Physics, University of Strathclyde, G4 0NG Glasgow, UK3 Leiden Observatory, Leiden University, PO Box 9513, 2300 RA Leiden, The Netherlands4 ESA European Space Research and Technology Centre (ESTEC), Keplerlaan 1, 2201 AZ Noordwijk, The Netherlands5 European Space Agency (ESA), European Space Astronomy Centre (ESAC), Villanueva de la Cañada, 28691 Madrid, Spain6 Institut für Astronomie und Astrophysik, Universität Tübingen, Sand 1, 72076 Tübingen, Germany

Received 26 February 2019 / Accepted 4 April 2019

ABSTRACT

Context. High-resolution X-ray spectroscopy has advanced our understanding of the hot Universe by revealing physical propertieslike kinematics, temperature, and abundances of the astrophysical plasmas. Despite technical and scientific achievements, the lackof scientific products at a level higher than count spectra is hampering complete scientific exploitation of high-quality data. Thispaper introduces the Catalog of Ionized Emission Lines Observed by the Reflection Grating Spectrometer (CIELO-RGS) onboard theXMM-Newton space observatory.Aims. The CIELO-RGS catalog aims to facilitate the exploitation of emission features in the public RGS spectra archive. In particular,we aim to analyze the relationship between X-ray spectral diagnostics parameters and measurements at other wavelengths. This paperfocuses on the methodology of catalog generation, describing the automated line-detection algorithm.Methods. A moderate sample (∼2400 observations) of high-quality RGS spectra available at XMM-Newton Science Archive is usedas our starting point. A list of potential emission lines is selected based on a multi-scale peak-detection algorithm in a uniform andautomated way without prior assumption on the underlying astrophysical model. The candidate line list is validated via spectral fittingwith simple continuum and line profile models. We also compare the catalog content with published literature results on a smallnumber of exemplary sources.Results. We generate a catalog of emission lines (1.2 × 104) detected in ∼1600 observations toward stars, X-ray binaries, supernovaeremnants, active galactic nuclei, and groups and clusters of galaxies. For each line, we report the observed wavelength, broadening,energy and photon flux, equivalent width, and so on.

Key words. X-rays: general – techniques: spectroscopic

1. Introduction

High-resolution X-ray spectroscopy has the potential to revealthe physics of the plasma state that accounts for much of thematerial in the observable universe. The X-ray band contains theinner-shell transitions of all abundant elements from C throughN, O, Ne, Mg, Si, S, Ca, Fe and Ni. X-ray spectra of cos-mic sources typically consist of a mixture of ionic or atomicemission or absorption lines and electron continuum compo-nents. Observed spectra can be modeled in order to identifythe underlying physics and dynamics of the plasma concerned.They often depart from the ideal conditions of equilibrium andspatial uniformity. It is not coincidental that the availability ofhigh-resolution data in the X-ray domain at the beginning of thecurrent millennium has triggered important developments in themodeling of X-ray emitting plasma (Behar et al. 2001; HitomiCollaboration 2018), in the collisionally ionized as well as in thephoto-ionized regime.

? The CIELO-RGS catalog is only available at the CDS via anony-mous ftp to cdsarc.u-strasbg.fr (130.79.128.5) or via http://cdsarc.u-strasbg.fr/viz-bin/qcat?J/A+A/625/A122

A giant leap in this field was achieved with the launchof Chandra and XMM-Newton carrying high-sensitivity, high-resolution grating systems. In particular, the Reflection Grat-ing Spectrometer (RGS; den Herder et al. 2001) on boardXMM-Newton (Jansen et al. 2001) consists of an array of reflec-tion gratings that diffract the light focused by two of the X-raytelescopes to an array of dedicated charge-coupled device detec-tors (CCDs). The RGS achieves high resolving power (150 to800) over an energy range from 0.35 to 2 keV ('6 to 35 Å, in thefirst spectral order). Thanks to the unprecedented collecting areaof the XMM-Newton telescopes, the RGS is the most sensitivehigh-resolution X-ray detector currently in operation. Its perfor-mance and excellent calibration quality (de Vries et al. 2015)have allowed for breakthrough advancements in astrophysicalfields as diverse as cool and active stars, X-ray binaries, galaxyclusters, and active galactic nuclei (AGNs), to mention just a few(Kahn et al. 2002).

Despite these outstanding technical and scientific achieve-ments, complete scientific exploitation of the plasma diagnosticinformation enclosed in RGS spectroscopic data is still partlyhampered by the lack of scientific products at a level higherthan count spectra. Astronomers interested in extracting plasmadiagnostics from RGS spectra must analyze them through the a

Article published by EDP Sciences A122, page 1 of 16

A&A 625, A122 (2019)

forward-folding technique, which requires specialized softwareand can be potentially cumbersome when applied to spectra withgood signal-to-noise ratios.

This paper describes CIELO-RGS1 (“CIELO” hereafter), acatalog of emission lines detected by the RGS instrument overmore than 15 years of XMM-Newton science operations. To ourknowledge CIELO is the first published catalog of its kind inhigh-energy astrophysics. The paper is organized as follows.Section 2 describes the procedure followed to reduce and selectthe optimal sample of RGS spectra from which the CIELOcatalog has been produced. Section 3 describes the algorithmemployed to select a sample of candidate emission lines on eachRGS spectrum, later validated through spectral fitting. The cat-alog structure is described in Sect. 4. A selection of prelimi-nary science results from the catalog as a whole is presented inSect. 5, followed by our conclusions in Sect. 6. Future paperswill discuss the application of CIELO to a wide range of astro-physical contexts and problems.

The CIELO catalog, data, and source codes used for creatingthe figures in this paper can be found at Zenodo2.

2. Spectra selection and reduction

The CIELO catalog is based on a subset of the RGS spectraavailable in the XMM-Newton Science Archive (XSA).

This subset has been taken from the work of Bensch et al.(2014). These authors examined all the RGS data available as of1 December 2014 in order to identify those that are scientificallyuseful.

The method devised was based on two properties: an esti-mation of the signal-to-noise ratio of the spectrum, for whicha minimum value of 10 was required, and a spatial profile ofthe spectrum along the cross-dispersion direction of the spectralimage. This profile was fitted with a Gaussian function. For thespectrum to be considered useful, the fitting parameters neededto be well determined and within some pre-defined range (e.g.,the width of the fitted Gaussian could not be narrower than theinstrumental spatial resolution, and the maximum had to be closeto the center of the field-of-view).

Out of the 8500 observations studied, 5000 (59%) passedthese two acceptance criteria. These 5000 observations were fur-ther sub-divided into two categories: “Top quality” observationsfor which both RGS 1 and RGS 2 spectra were selected (28% ofthe initial sample), and “Good” observations, with only RGS 1or RGS 2 considered as useful (31% of the initial sample). Thesample used here is composed of the 2421 observations classi-fied as Top quality in Bensch et al. (2014).

The data used in this work were taken from the XMM-NewtonScience Archive (XSA). For each of the selected observations, thefollowing data were downloaded: extracted first- order total (i.e.,source+background) spectra, extracted first-order backgroundspectra, and first-order response matrices (those for RGS alsoinclude the effective area), for RGS 1 and RGS 2.

These data were obtained from the raw observation data file(ODF) after automatic processing with the pipeline processingsystem (PPS). The PPS extracts RGS source spectra using thecoordinates given in the original observational proposal, andan extraction region with a spatial size of 95% of the cross-dispersion PSF. The background is extracted from the regionbetween 98% and 100% of the cross-dispersion PSF. A regionincluding 95% of the pulse-height distribution is used in theextraction of both source and background spectra.1 Catalog of Ionized Emission Lines Observed by the RGS.2 http://doi.org/10.5281/zenodo.2577859

3. Catalog generation

Sophisticated plasma models are widely used for detailed anal-ysis of emission features in high-resolution X-ray spectra (e.g.,Broersen et al. 2011; Gu et al. 2016; Mao et al. 2018, 2019).While this approach yields self-consistent physical interpreta-tions (e.g., temperature, density, and abundance) of the spec-tral features, it requires extensive prior knowledge of the sourceof interest, including but not limited to the cosmological red-shift, line-of-sight galactic absorption, which and how manyplasma models to use (collisional ionized, photoionized, non-equilibrium ionization, and charge exchange), and sometimes amulti-component model of the continuum. This approach is notapplicable for the analysis of a large sample like CIELO-RGS.

An alternative approach is to use a phenomenological fit (e.g.,Guainazzi & Bianchi 2007; Pinto et al. 2016; Psaradaki et al. 2018)for individual emission features. Typically, the (local) continuumis simply modeled as a (local) power law or spline function. Subse-quently, it is common for the investigators to identify the emissionfeatures by eye or to scan the spectrum using a Gaussian/delta pro-file. Each emission feature is then modeled with a delta functionor a Gaussian function with fixed width.

The phenomenological analysis is straightforward. Itrequires no prior knowledge of the source if all the emissionlines are narrow and unresolved. When the emission lines arebroader than the spectral resolution of the instrument, assumingthat all the emission features have identical broadening, then theonly prior knowledge needed is that of the line broadening.

The above issue with conventional phenomenological analy-sis is tractable with a line-detection algorithm based on wavelettransform. We note that wavelet transform analyses have provento be suitable for both spectral and imaging analysis in astronomy(e.g., Bury et al. 1996; Starck et al. 1997; Machado et al. 2013). Inparticular, the concept of the multi-scale peak detection (MSPD)initially proposed in the chemistry community Zhang et al. (2015,Z15 hereafter) was followed for the present work.

The advantage of MSPD is that it extracts information fromhigh-throughput spectra in a rapid way without any prior knowl-edge. Strong and weak, broad and narrow, symmetric and asym-metric spectral features on top of a continuum can be easilydetected (Sect. 3.1). Once the number of significant spectral fea-tures is determined, the aforementioned phenomenological fitcan be performed in a uniform and automated way (Sect. 3.2).

We should point out that the original MSPD algorithm shownin Z15 needs to be tailored for RGS data analysis. Z15 tookadvantage of “ridges” (local maxima, explained in Sect. 3.1),“valleys” (local minima) and “zero-crossings” (sign changes)for the peak detection. As MSPD does not estimate/fit the con-tinuum, the presence of continuum breaks in the RGS spectra(e.g., CCD gaps) would lead to artificial “valleys” and “zero-crossings”. Furthermore, in Z15, the signal and noise levels aredefined as the maximum and the 90th percentile within a selectedwindow for each potential peak3. This would not work properlyfor closely located emission features, such as for example He-like triplets in many astrophysical spectra.

3.1. Multi-scale peak detection

The present MSPD algorithm flows as follows. It is based onthe continuous wavelet transform (CWT) using the Mexican hat

3 Details can be found in the source code available athttps://github.com/zmzhang/libPeak/blob/master/peak_detection.py

A122, page 2 of 16

J. Mao et al.: CIELO-RGS

−100 −50 0 50 100

x

−0.4

−0.2

0.0

0.2

0.4

0.6

0.8

1.0

ψ(x,s

)

scale = 1scale = 5scale = 10scale = 20

Fig. 1. Mexican hat wavelet at different scales. Here we use 201 points(i.e., 200 bins) to describe the wavelet. The corresponding Y-value isplaced at the center of the each bin to mimic the real spectra.

wavelet (or Ricker wavelet)

ψ(x, s) =2

√3 sπ1/4

(1 −

x2

s2

)exp

(−

x2

2 s2

), (1)

where x is the number of points used to describe the wavelet ands is the scale of the wavelet. Figure 1 shows an example of theMexican hat wavelet at some scales.

The input spectrum S (N) is convolved with a set of waveletfunctions ψ(N′, s), where N′ is the minimum value between thenumber of bins (N) in the input spectrum and ten times the scale(10 × s) and the scale (s) ranges from 1 to 50. The top panelof Fig. 2 shows the observed RGS flux spectrum of NGC 1068(ObsID: 0111200101) in the 6−8 Å wavelength range (∼3200spectral bins). The second panel of Fig. 2 presents the results ofthe continuous wavelet transform (CWT) with s = 3 in blue and37 in orange, respectively. Peaks appear in the observed spec-trum yield larger CWT coefficients. The third panel of Fig. 2presents the color-coded CWT coefficients. This can be viewedas the dilation and translation of peaks. The redder (larger) thecoefficient, the more prominent the peak. Ridges are introducedas curves of local maxima in the 2D wavelet space, where therows and columns are scales and bins, respectively. The bot-tom panel of Fig. 2 illustrates ridges across different scales forNGC 1068. Ridges are denser at smaller scales as the spectrumis nosier.

A ridge in the wavelet space does not necessarily correspondto a prominent emission feature in the observed spectrum (e.g.,Fig. 3). Starting from the left-hand side of the 2D wavelet space(i.e., shorter observed wavelength), for each ridge, we track thebin numbers and CWT coefficients from the largest scale presentdown to s = 2. We note that a ridge starting from a larger scalemight diverge into two or more ridges at smaller scales due tothe presence of noise. Therefore, we stop tracking the ridges ats = 2. The ridge of a potential line should meet the followingempirical criteria:1. Local maxima should appear at larger scales s & 6;2. At least 20% of bin numbers should be identical so that the

ridge is relatively straight;3. The maximum CWT coefficient should be at least 2.7σ

above the average.The quoted numbers are used for the present sample analysis,which can certainly be fine tuned for a case study. The wave-length corresponds to the bin number appearing in most scalesof the selected ridge that is assigned to the potential line. This

choice of wavelength is arbitrary but not critical because it willbe determined more accurately via spectral fitting (Sect. 3.2).

While the present MSPD algorithm successfully detects themajority of visible peaks (by eye) in the point-like sourceNGC 1068, we need to validate its performance on extendedsources. Since the RGS spectrometer is slitless, the emissionline profiles of extended sources would appear broader, whichcomplicates the analysis in practise. Figure 4 shows the applica-tion of the present MSPD algorithm to the observed flux spec-trum of NGC 5044 (ObsID: 0554680101). Again, the majorityof visible peaks (by eye) are detected. Meanwhile, the rela-tively poorer signal-to-noise of the spectrum leads to more falsedetections. These false peaks are filtered out after the followingspectral fitting.

3.2. Spectral fitting

The first-order background subtracted RGS 1 and RGS 2 spec-tra, as well as response matrices, are converted into the SPEX(Kaastra et al. 1996, 2018) format through the SPEX task trafo.For each observation, RGS 1 and RGS 2 spectra, if available, arefitted simultaneously. About 1% of the observations in the BIRDcatalog have either only RGS 1 or RGS 2 data, which are alsofitted with the data available.

The full-width-half-maximum (FWHMinst) of the first-orderRGS instruments is ∼0.06−0.09 Å4. For extended sources, dueto the spatial broadening (θ ≥ 0.8 arcmin or 90% of the point-spread-function) along the cross-dispersion direction, the effec-tive FWHMeff is bigger than the instrumental FWHMinst,

FWHMeff =0.138

mθ

arcminÅ, (2)

where m is the spectral order (Tamura et al. 2004) and, in ourcase, m = 1.

The default bin size of the RGS spectra is 0.01 Å, which isclearly over-sampling. Therefore, we re-bin each RGS spectrumto achieve optimal binning (Kaastra & Bleeker 2016), which isone half to one third of the maximum between FWHMinst andFWHMeff . That is to say, in any case, the minimum binning fac-tor is 3.

Wavelength and flux values in CIELO are observed measure-ments. We did not apply any corrections for redshift or for X-rayabsorption.

For each spectrum, the continuum is modeled with a splinefunction, which consists of a grid of nine points evenly distributedin the 6−38 Å wavelength range. The spline continuum modelinghas been used successfully in for example Kaastra et al. (2011).

Without any prior knowledge of the astrophysical propertiesof the targets, all emission features are modeled with Gaussianprofiles. The multi-scale peak-detection algorithm (Sect. 3.1)provides the number of emission features to SPEX, as well asthe initial guess of the wavelength of each feature. The normal-ization, wavelength, and broadening of each Gaussian are free tovary during the fit. We note that the velocity broadening of eachGaussian is constrained to be 0−104 km s−1. The upper limit isset to cover broad emission features from the broad-line regionin AGNs (Blandford et al. 1990).

The optimization method used here is the classical“Levenberg-Marquardt” method, which might not lead to theglobal minimum. A better fit can be found during the error search

4 We refer readers to Fig. 82 of the XMM User’s hand book(available at https://xmm-tools.cosmos.esa.int/external/xmm_user_support/documentation/uhb/rgsresolve.html) fordetails.

A122, page 3 of 16

A&A 625, A122 (2019)

020406080

100

Coun

ts s

−1 m

−2 Å

−1

NGC 1068 (0111200101)

−50

0

50

100

CWT

coef

ficie

nts

scale = 3scale = 37

1

10

20

30

40

50

Scal

es

CWT coefficients

1 500 1000 1500 2000 2500 3000Bin

1

10

20

30

40

50

Scal

es

Ridges

10 15 20 25 30 35Observed wavelength (Å)

Fig. 2. Top panel: observed RGS flux spectrum of NGC 1068 (ObsID: 0111200101). The RGS 1 and RGS 2 spectra (in red) are combined. Errorbars are not shown for clarity. The blue triangles mark peaks detected by the MSPD algorithm (Zhang et al. 2015) with (their definition of) signal-to-noise ratio ≥7. The yellow circles mark peaks detected by the present work at a significance level of &2.7σ. The estimated continuum in greenis based on the asymmetrically reweighted penalized least squares algorithm (arPLS, Baek et al. 2015). The estimated continuum is shown merelyfor guidance as it is independent of the MSPD algorithm. The second panel gives the coefficients (Y-axis) of the continuous wavelet transform(CWT) of the spectrum at scale = 3 (blue) and 37 (orange). The third panel is a color map of the CWT coefficients at different scales (Y-axis) andat each spectral bin (X-axis). The redder the color, the larger the CWT coefficient. The bottom panel presents the so-called ridges across differentscales.

of the parameters for each Gaussian. If so, we refit the spectrumstarting from those values that lead to better (or lower) statistics.We stop the iteration if the improvement on the ∆C . 0.5, andlimit the number of iterations (≤5) if the convergence is slow.The slow convergence is partly due to those very broad emissionfeatures, where the broadening and the normalization parametersare degenerate to some extent.

The best-fit parameters of an emission line are included inthe final catalog if its normalization is above zero at a 3σ confi-dence level and the energy flux is &10−18 W m−2.

4. Catalog structure

The CIELO catalog is in the form of a machine readable ASCIItable and a FITS table. Both tables consist of 10981 rows from

1492 unique observations and 16 columns: Col. (1) is the ObsID;Cols. (2)−(4) are the wavelength (in Å) and the 1σ lower andupper limits; Cols. (5)−(7) are the FWHM in Å and the 1σlower and upper limits; Cols. (8)−(10) are normalization (in1044 ph s−1) and the 1σ lower and upper limits; Col. (11) is theobserved energy flux (in W m−2); Col. (12) is the observed pho-ton flux (in ph m−2 s−1); Col. (13) is the modeled (spline) contin-uum photon flux density at the line center (in ph m−2 s−1 Å−1),and Col. (14) is the equivalent width (EW in Å) defined asfollows:

EW =Fline

Fcont, (3)

where Fline is the observed photon flux of the line (Col. 12) andFcont is the modeled (spline) continuum photon flux density at

A122, page 4 of 16

J. Mao et al.: CIELO-RGS

0

20

40

60

80

100

Coun

ts s

−1 m

−2 Å

−1

NGC 1068 (0111200101)

−50

0

50

100

CWT

coef

ficie

nts

scale = 3scale = 37

1

10

20

30

40

50

Scal

es

CWT coefficients

1500Bin

1

10

20

30

40

50

Scal

es

Ridges

18.5 19.0 19.5 20.0 20.5 21.0 21.5 22.0 22.5Observed wavelength (Å)

Fig. 3. Similar to Fig. 2 but in the neighborhood of the O viii Lyα line and the O vii He-like triplets.

the line center (Col. 13). The last two columns are the redshift(from Simbad, Col. 15) and target name (from XSA, Col. 16).

5. Results

CIELO includes data from ∼66% of the BIRD “top quality”observations. The targets of the remaining observations includesome pulsars, BL-Lac objects, gamma-ray bursts, voids, and soon. The soft X-ray spectra of these targets are featureless withsmall uncertainties on the (continuum) flux.

Figure 5 gives an overview of the observed wavelength distri-bution of all the lines included in the catalog. H-like Lyman series,He-like triplets, and Ne-like Fe xvii resonance, and forbidden linesdominate the catalog. Since observations are biased to nearby andbright targets, the observed wavelength of the emission lines peaksaround the corresponding rest frame wavelength.

Here we present comparisons of our results with published lit-erature results. The comparisons are limited to point-like sources.As for extended sources like groups and clusters of galaxies(e.g., Kaastra et al. 2001; Tamura et al. 2001; Werner et al. 2006)

and supernovae remnants (e.g., Rasmussen et al. 2001; van derHeyden et al. 2003; Broersen et al. 2011), results of plasma mod-eling (e.g., temperature, abundance, emission measure) are usu-ally provided, instead of a list of emission lines.

5.1. Local obscured AGN: NGC 1068

NGC 1068 is the prototypical Seyfert 2 galaxy in the local Uni-verse with a redshift of z = 0.0038 (Huchra et al. 1999). Herewe take the RGS spectrum of NGC 1068 (ObsID: 0111200101,41 ks effective exposure) as an example for local obscuredAGNs. We show in Fig. 6 a comparison of the emission linesdetected in the present study and Kinkhabwala et al. (2002, K02hereafter). More details can be found in Table A.1. We also showin Fig. A.1 the results of spectral fitting (Sect. 3.2).

K02 reported measurements of 40 lines and six RRC in theirTables 1 and 2, respectively. The CIELO catalog contains 38lines (including RRC). The difference in the total number oflines is not unexpected. The present analysis does not bene-fit from prior knowledge of the atomic origin of the emission

A122, page 5 of 16

A&A 625, A122 (2019)

0

5

10

Coun

ts s

−1 m

−2 Å

−1

NGC 5044 (0554680101)

−10

−5

0

5

10

CWT

coef

ficie

nts

scale = 11scale = 43

1

10

20

30

40

50

Scal

es

CWT coefficients

1 500 1000 1500 2000 2500 3000Bin

1

10

20

30

40

50

Scal

es

Ridges

10 15 20 25 30 35Observed wavelength (Å)

Fig. 4. Similar to Fig. 2 but for NGC 5044 (ObsID: 0554680101).

features. Consequently, the resonance and intercombinationlines of Ne ix, for example, are fitted as two lines in K02, yet asone line (i.e., a single Gaussian profile) here; the O viii Lyγ andputative Fe xxiv, Ar xi, and S xii lines are missed in K02 but areincluded here. Moreover, the present work is designed (Sect. 3.1)to omit relatively weak lines. Accordingly, weak lines like theO viii Lyδ line are included in K02 but are missed here.

If lines are well separated, the velocity shifts (voff) and broad-ening (σv), and the observed photon fluxes (Fpho) from the abovetwo analyses are consistent with each other. The Ne x Lyα lineis an exception, where the observed wavelength and velocitybroadening are (12.180 ± 0.004) Å and 470 ± 120 km s−1 in K02yet (12.206 ± 0.005) Å and 1200 ± 160 km s−1 in the presentwork. The observed Ne x Lyα line profile shown in Fig. A.1 devi-ates to a Gaussian profile, which leads to the measurement bias.We also note that 12.165 Å and 1400+200

−14 km s−1 are quoted byKallman et al. (2014, K14 hereafter), where the spectral analysisis based on the 450 ks Chandra high-energy transmission gratingspectroscopy data.

If lines are blended with each other, disagreement on thevelocity shift and broadening are expected. The Ne ix He-liketriplets around 13.5 Å is a typical example. We note that both thepresent work and K14 cannot distinguish the resonance from theintercombination lines.

Furthermore, at shorter wavelengths (λ . 20 Å), the linewidths obtained in the present work are in general larger thanthose given in K02, but within the upper limits given by K14.Better agreement is found at longer wavelengths (λ & 20 Å).

There are two caveats in Table A.1. First, K02 assign a singlevalue of 6.69 Å to the rest-frame wavelength of the Si xiiiHe-liketriplets. The wavelengths of the resonance, intercombination,and forbidden lines are 6.648 Å (w), 6.685 Å (x), 6.688 Å (y), and6.739 Å (z), respectively. Second, the present work fails to prop-erly interpret the intercombination lines (x and y) of N vi around29 Å. The line profile has a single peak in RGS 1 but a doublepeak in RGS 2 (the fourth panel of Fig. A.1). This leads to sig-nificantly biased results. K02 tied the velocity broadening of theHe-like triplets and yielded more proper results.

A122, page 6 of 16

J. Mao et al.: CIELO-RGS

10 15 20 25 30 35Observed wavelength (Å)

0

100

200

300

400

500

600

700

Numbe

r of lines

Si XIII

Mg XII

Mg XI

Ne X Ne IX

Fe XVII

Fe XVIII

Fe XVII

O VIII

O VII

N VII

N VI

C VI

Fig. 5. Histogram of the CIELO catalog as a function of observed wave-length. The rest frame wavelength of H-like Lyα and He-like resonanceslines are labeled for guidance.

5.2. Stars: Procyon and HD 159176

High-quality RGS spectra of various active stars are available.We take Procyon and HD 159176 as two examples to illustrateresults from the present analysis.

The X-ray spectrum of Procyon is dominated by the late-type(F5 IV-V) optically bright star (with a faint white dwarf compan-ion). The stellar corona has a temperature of ∼(1−3)×106 K andexhibits a cooler X-ray spectrum than magnetically active stars(Raassen et al. 2002, R02 hereafter). In Fig. 7, we compare theobserved wavelengths and fluxes of ObsID 0123940101 (45.2 kseffective exposure) with those from Table 1 of R02, where bothRGS and LEGS results are available. More details can be foundin Table A.2. We also show in Fig. A.2 the results of spectralfitting (Sect. 3.2).

Apart from those lines that we fail to individually identifyhere (e.g., around 13.56 Å and 36.45 Å), a good agreement canbe found between the present work and R02 in terms of theobserved wavelength. As for the line (photon) fluxes, we findlarger values in the present work in addition to only a few con-sistent results (e.g., the O viii 18.628 line). This is mainly due tothe two different approaches for the continuum. As described inSect. 3.2, a spline function is used here for the global continuum.On the other hand, R02 adopted a constant local “background”level. The local background was adjusted for each emission lineto account for the real continuum and the pseudo-continuum cre-ated by the overlap of several neglected weak lines.

Unlike the narrow and “regular” emission lines in Procyon,the emission lines in HD 159176 are rather broad and asymmet-ric (neither Gaussian nor Lorentzian). HD 159176 is a double-lined spectroscopic binary (O7V + O7V) in the young opencluster NGC 6383 (Trumpler 1930). As can be seen in Fig. A.3,the Gaussian profiles adopted in the present analysis manage tocapture the main features yet fail to account for the details. Wealso note that fitting RGS 1 and RGS 2 simultaneously rejectsthose features that appear in only one of the instruments (e.g.,around ∼10 Å) and those features that appear only in one instru-ment (e.g., the one around ∼9.3 Å).

According to De Becker et al. (2004), the O viii Lyαline has an estimated FWHM of 2500 km s−1. Moreover, it is

slightly blue-shifted ∼300−600 km s−1, though the highest peakin the profile is found near the rest frame wavelength 18.967 Å(De Becker et al. 2004). The present analysis yields an FWHMof (0.130 ± 0.014) Å, i.e., (2060 ± 220) km s−1 and an observedwavelength of (18.975± 0.008) Å, that is, blue-shifted by (130±120) km s−1. Given the complicated nature of this spectrum, ourresults are acceptable.

5.3. X-ray binaries: Her X-1

Her X-1 is a bright intermediate-mass X-ray binary (Schreieret al. 1972) exhibiting an unusual long-term X-ray flux modula-tion with a period of 35 days (Giacconi et al. 1973). The 35-dayX-ray light curve is asymmetric and contains two peaks: a state ofeight-day duration reaching the peak flux Fmax named the main-on state and a four-day duration reaching one third of the max-imum named the short-on state. The rest is named the low statewith flux twenty times lower than the maximum (Jimenez-Garateet al. 2002, JG02).

We inspect the RGS spectra of Her X-1 at three differ-ent orbital phases: 0134120101 (main), 0111060101 (low), and0111061201 (short-on). Figure 8 visualizes the comparison ofemission lines reported in the present work and JG02. Tabulatedresults are given in Table A.3. The plot of the spectral fitting forthe short-on phase can be found in Fig. A.4.

Minor differences are found between the present analysis andJG02. The present work does not include the weak resonanceline in the O vii He-like triplets. The N vii Lyα line is found to berather broad (4200±1600) km s−1 in the present work. The widthwas fixed to 3200 km s−1 in JG02, based on their width of O viiHe-like triplets. As a result, the line (photon) flux reported hereis slightly higher (at a 2σ confidence level) than that of JG02.

Moreover, we also include a few broad features at 6.4 Å,6.7 Å, and 14.1 Å. Such broad features are likely due to calibra-tion uncertainties of the instrument. The smaller the statisticaluncertainty, the more prominent such systematic uncertainties.

5.4. Comparison of plasma diagnostics: stars vs. obscuredAGNs

The intensity of the brightest components of n = 2 to n = 1 tran-sitions in He-like ions can be combined into ratios with a strongdiagnostic power of the electron density [R(ne)] and electronictemperature [G(Te)] (Gabriel & Jordan 1969):

R(ne) =fi

(4)

G(Te) =f + i

r, (5)

where f , i, and r are the intensities of the forbidden (z; 1s2 1S0– 1s2s 3S1), intercombination (x,y; 1s2 1S0 – 1s2p 3P2,1), andresonance (w; 1s2 1S0 – 1s2p 1P1) lines, respectively. The samediagnostic ratios can be used to study photoionised plasma aswell (Porter & Ferland 2007). Indeed, they can be used toidentify the dominant ionization mechanism (Porquet & Dubau2000).

The CIELO catalog allows for studies of these diagnos-tic parameters on large source populations. As an example, inFig. 9 we show the R versus G diagnostic plot for the OVIIHe-α transition derived from the CIELO catalog for the twosoft, X-ray brightest, heavily obscured AGNs (NGC 1068, andNGC 4151) (Guainazzi & Bianchi 2007), and for a sample ofactive stars. Multiple data points for each source correspond

A122, page 7 of 16

A&A 625, A122 (2019)

5 10 15 20 25 30 35Observed wavelength (Å)

10−1

100

101Fl

ux (p

h m

−2 s

−1)

Present work

Kinkhabwala+2002

Fig. 6. Emission lines detected in the RGS spectrum of NGC 1068 during the present study (black) and by Kinkhabwala et al. (2002, red). Forvisibility, for each data point we assign the line width (in Å instead of the wavelength uncertainty) to the horizontal error bar.

15 20 25 30 35Observed wavelength (Å)

0.0

1.0

2.0

3.0

4.0

5.0

6.0

Flux

(ph

m−2

s−1

)

Present work

Raassen+2002 (RGS1)

Raassen+2002 (RGS2)

Raassen+2002 (LETGS)

Fig. 7. Emission lines detected in the RGS spectrum of Procyon during the present study (black) and Raassen et al. (2002, red for RGS 1, blue forRGS 2, and orange for LETGS). The wavelength uncertainty is not visible here.

to different observations. The AGN X-ray spectra are domi-nated by plasma photoionized by the intense radiation field ofthe active nucleus (Kinkhabwala et al. 2002), most likely dueto X-ray diffuse emission associated with the extended nar-row line regions (Bianchi et al. 2006). The variability of the Rparameter reflects time changes in the ratio between the forbid-den and the intercombination component that could be relatedto the UV illumination of the ionized nebula in a radiation-driven wind (Landt et al. 2015). Stellar spectra should be insteaddominated by optically thin collisionally ionized plasma. TheCIELO diagnostic diagram nicely separates the two classes, asexpected.

6. Summary

In this paper we present CIELO-RGS, the first systematic cat-alog of emission lines detected by the RGS instrument (denHerder et al. 2001) on-board XMM-Newton. CIELO-RGS wasbuilt using high-quality spectra accumulated during more than5000 observations performed over 15 years of XMM-Newtonscience operations. This catalog is intended to be an additionaltool to facilitate the exploitation of the RGS spectra available in

the public science archive, and in particular for comparisons ofX-ray spectral diagnostic parameters and measurements at otherwavelengths.

The catalog generation is based on a two-tier process. First,each RGS spectrum was analyzed searching for candidate emis-sion lines with a technique fully independent of the spectralmodeling, and therefore in principle of the intrinsic physical andastrophysical nature of the emitting plasma. This technique isone version of a multi-scale peak-detection algorithm, robustlytested in other scientific contexts (Z15). This step provides a listof candidate emission lines detected in each individual obser-vation spectrum over the appropriate range of frequency scales(upper-bound by the instrumental resolution). This list of candi-date emission lines is subsequently validated through a formalforward-folding spectral fitting process that properly takes intoaccount the instrumental transfer function, the observation back-ground, and the rigorous statistical conditions for spectral bin-ning (Kaastra & Bleeker 2016). The resulting catalog includesthe following measured observable for each line: wavelength,intrinsic profile width, normalization, energy and photon flux,and equivalent width, as well as a phenomenological parameter-ization of the underlying continuum.

A122, page 8 of 16

J. Mao et al.: CIELO-RGS

4

6

8

10

12

14

16

18

Flux

(ph

m−2

s−1

)

Phase main (a)

Present workJimenez-Garate+2002

0

1

2

3

4

Flux

(ph

m−2

s−1

)

Phase low (b)

Present workJimenez-Garate+2002

10 15 20 25 30 35Observed wavelength (Å)

0

1

2

3

4

5

6

7

8

Flux

(ph

m−2

s−1

)

Phase short-on (c)

Present workJimenez-Garate+2002

Fig. 8. Comparison of emission lines detected in the RGS spectra of HerX-1 in three different orbital phases. The present work and Jimenez-Garate et al. (2002) are shown in black and red, respectively.

Fig. 9. OVII He-α diagnostic parameters R and G for the two brightestheavily obscured AGNs (NGC 1068, blue; NGC 4151, red), and for asample of active stars (green) in CIELO-RGS.

In this paper we present some illustrative examples of thewide range of scientific themes that such a catalog can address,such as the ionization mechanism and dynamics of the extendednarrow line region in AGNs, plasma diagnostics in optically thin,collisionally ionised plasma in stars or the intra-cluster mediumpermeating galaxy clusters, and coronal emission in accreting

stellar-mass black holes. These topics will be further developedin specific future papers.

Acknowledgements. We thank the referee for careful reading of the manuscript.The authors gratefully acknowledge a generous financial support by the Divi-sional Funding of the Science Support Office at ESA, in the framework of theEXPRO contract 40000124950/18/NL/JB/gg. J. M. acknowledges discussionsand consultations with J. de Plaa, N. Loiseau, M. Mehdipour, and C. de Vries. JMis supported by STFC (UK) through the University of Strathclyde UK APAP net-work grant ST/R000743/1. SRON is supported financially by NWO, the Nether-lands Organization for Scientific Research.

ReferencesBaek, S.-J., Park, A., Ahn, Y.-J., & Choo, J. 2015, Analyst, 140, 250Behar, E., Sako, M., & Kahn, S. M. 2001, ApJ, 563, 497Bensch, K., Santos-Lleo, M., & Gonzalez-Riestra, R. 2014, The X-ray Universe

2014, 36Bianchi, S., Guainazzi, M., & Chiaberge, M. 2006, A&A, 448, 499Blandford, R. D., Netzer, H., Woltjer, L., Courvoisier, T. J. L., & Mayor, M. 1990,

Active Galactic Nuclei (New York: Springer-Verlag Berlin Heidelberg), 97Broersen, S., Vink, J., Kaastra, J., & Raymond, J. 2011, A&A, 535, A11Bury, P., Ennode, N., Petit, J.-M., et al. 1996, NIMPA, 383, 572De Becker, M., Rauw, G., Pittard, J. M., et al. 2004, A&A, 416, 221den Herder, J. W., Brinkman, A. C., Kahn, S. M., et al. 2001, A&A, 365, L7de Vries, C. P., den Herder, J. W., Gabriel, C., et al. 2015, A&A, 573, A128Gabriel, A. H., & Jordan, C. 1969, Nature, 221, 947Giacconi, R., Gursky, H., Kellogg, E., et al. 1973, ApJ, 184, 227Gu, L., Kaastra, J., & Raassen, A. J. J. 2016, A&A, 588, A52Guainazzi, M., & Bianchi, S. 2007, MNRAS, 374, 1290Hitomi Collaboration (Aharonian, F., et al.) 2018, PASJ, 70, 12Huchra, J. P., Vogeley, M. S., & Geller, M. J. 1999, ApJS, 121, 287Jansen, F., Lumb, D., Altieri, B., et al. 2001, A&A, 365, L1Jimenez-Garate, M. A., Hailey, C. J., den Herder, J. W., Zane, S., & Ramsay, G.

2002, ApJ, 578, 391Kaastra, J. S. 2017, A&A, 605, A51Kaastra, J. S., & Bleeker, J. A. M. 2016, A&A, 587, A151Kaastra, J. S., Mewe, R., & Nieuwenhuijzen, H. 1996, UV and X-ray

Spectroscopy of Astrophysical and Laboratory Plasmas (Tokyo: Univ. Ac.Press), 411

Kaastra, J. S., Ferrigno, C., Tamura, T., et al. 2001, A&A, 365, L99Kaastra, J. S., Steenbrugge, K. C., Detmers, R. G., et al. 2011, A&A, 534, A38Kaastra, J. S., Raassen, A. J. J., de Plaa, J., & Gu, L. 2018, SPEX X-ray Spectral

Fitting Package (Version 3.05.00). Zenodo, DOI: 10.5281/zenodo.2419563Kahn, S. M., Behar, E., Kinkhabwala, A., & Savin, D. W. 2002, RSPTA, 360,

1923Kallman, T., Evans, D. A., Marshall, H., et al. 2014, ApJ, 780, 121Kinkhabwala, A., Sako, M., Behar, E., et al. 2002, ApJ, 575, 732Landt, H., Ward, M. J., Steenbrugge, K. C., & Ferland, G. J. 2015, MNRAS,

449, 3795Machado, D. P., Leonard, A., Starck, J.-L., Abdalla, F. B., & Jouvel, S. 2013,

A&A, 560, A83Mao, J., Kaastra, J. S., Mehdipour, M., et al. 2018, A&A, 612, A18Mao, J., de Plaa, J., Kaastra, J. S., et al. 2019, A&A, 621, A9Pinto, C., Middleton, M. J., & Fabian, A. C. 2016, Nature, 533, 64Porquet, D., & Dubau, J. 2000, A&AS, 143, 495Porter, R. L., & Ferland, G. J. 2007, ApJ, 664, 586Psaradaki, I., Costantini, E., Mehdipour, M., & Díaz Trigo, M. 2018, A&A, 620,

A129Raassen, A. J. J., Mewe, R., Audard, M., et al. 2002, A&A, 389, 228Rasmussen, A. P., Behar, E., Kahn, S. M., den Herder, J. W., & van der Heyden,

K. 2001, A&A, 365, L231Schreier, E., Levinson, R., Gursky, H., et al. 1972, ApJ, 172, L79Starck, J.-L., Siebenmorgen, R., & Gredel, R. 1997, ApJ, 482, 1011Tamura, T., Kaastra, J. S., Peterson, J. R., et al. 2001, A&A, 365, L87Tamura, T., Kaastra, J. S., den Herder, J. W. A., Bleeker, J. A. M., & Peterson, J.

R. 2004, A&A, 420, 135Trumpler, R. J. 1930, PASP, 42, 342van der Heyden, K. J., Bleeker, J. A. M., Kaastra, J. S., & Vink, J. 2003, A&A,

406, 141Werner, N., Böhringer, H., Kaastra, J. S., et al. 2006, A&A, 459, 353Zhang, Z.-M., Tong, X., Peng, Y., et al. 2015, Analyst, 140, 7955

A122, page 9 of 16

A&A 625, A122 (2019)

Appendix A: Detailed comparisons of the examples

We show the best-fit models against the observed spectrafor NGC 1068, Procyon, HD 15176, and Her X-1 in Sect. 5

(Figs. A.1–A.4). Comparisons of the emission lines among thepresent work and others are also shown (Tables A.1–A.3).

6.0 6.5 7.0 7.5 8.0 8.5 9.0 9.5 10.0

0.000

0.015

0.030

Counts/s/Å

ObsID: 0111200101 (rebin by: 3)

10 11 12 13 14 15 16 170.00

0.03

0.06

0.09

0.12

Counts/s/Å

17 18 19 20 21 22 23 24

0.0

0.1

0.2

0.3

Counts/s/Å

24 25 26 27 28 29 30 31

0.00

0.06

0.12

Counts/s/Å

31 32 33 34 35 36 37 38

Observed wavelength (Å)

0.00

0.06

0.12

Counts/s/Å

Fig. A.1. RGS spectrum of NGC 1068 (ObsID: 0111200101). RGS 1 and RGS 2 spectra are fitted simultaneously. RGS spectra are rebinned by afactor of 3. RGS 1 and RGS 2 data are shown in black and gray, respectively. The best-fit models are shown in red and pink for RGS 1 and RGS 2,respectively. The breaks (e.g., around 11.2 Å and 24 Å) are due to CCD gaps, bad pixels, etc.

A122, page 10 of 16

J. Mao et al.: CIELO-RGS

6.0 6.5 7.0 7.5 8.0 8.5 9.0 9.5 10.0−0.010

−0.005

0.000

0.005

0.010

Counts

/s/Å

ObsID: 0123940101 (rebin by: 3)

10 11 12 13 14 15 16 17−0.01

0.00

0.01

0.02

Counts

/s/Å

17 18 19 20 21 22 23 24

0.00

0.06

0.12

Counts

/s/Å

24 25 26 27 28 29 30 31

0.000

0.025

0.050

Counts

/s/Å

31 32 33 34 35 36 37 38

Observed wavelength (Å)

0.00

0.05

0.10

Counts

/s/Å

Fig. A.2. Similar to Fig. A.2 but for Procyon (ObsID: 0123940101).

A122, page 11 of 16

A&A 625, A122 (2019)

6.0 6.5 7.0 7.5 8.0 8.5 9.0 9.5 10.0 10.5

−0.02

0.00

0.02

0.04

0.06

0.08

Counts

/s/Å

ObsID: 0001730201 (rebin by: 3)

10 11 12 13 14 15 16 17

−0.02

0.00

0.02

0.04

0.06

0.08

Counts

/s/Å

16 17 18 19 20 21 22 23 24

−0.02

0.00

0.02

0.04

0.06

0.08

Counts

/s/Å

23 24 25 26 27 28 29 30 31

−0.02

0.00

0.02

0.04

0.06

0.08

Counts

/s/Å

31 32 33 34 35 36 37 38

Observed wavelength (Å)

−0.02

0.00

0.02

0.04

0.06

0.08

Counts

/s/Å

Fig. A.3. Similar to Fig. A.1 but for HD 159176 (ObsID: 0001730201).

A122, page 12 of 16

J. Mao et al.: CIELO-RGS

6.0 6.5 7.0 7.5 8.0 8.5 9.0 9.5 10.0

0.00

0.04

0.08

Counts/s/Å

ObsID: 0111061201 (rebin by: 3)

10 11 12 13 14 15 16 17

0.00

0.05

0.10

Counts/s/Å

17 18 19 20 21 22 23 240.00

0.08

0.16

0.24

Counts/s/Å

24 25 26 27 28 29 30 31

0.00

0.08

0.16

Counts/s/Å

31 32 33 34 35 36 37 38

Observed wavelength (Å)

0.00

0.03

0.06

Counts/s/Å

Fig. A.4. Similar to Fig. A.1 but for Her X-1 (ObsID: 0111061201).

A122, page 13 of 16

A&A 625, A122 (2019)

Table A.1. List of lines and radiative recombination continua (RRC) in NGC 1068.

Line ID λ0 voff σv Fpho λobs voff σv Fpho

Å km s−1 km s−1 ph m−2 s−1 Å km s−1 km s−1 ph m−2 s−1

Si xiv Lyα 6.182 . . . . . . .0.8 ± 0.2Si xiii 6.69 . . . . . . .0.8 ± 0.2 6.74† . . . . . . 0.77 ± 0.08Si i K 7.154 . . . . . . .0.8 ± 0.2Mg xii Lyα 8.421 70 ± 250 . . . 0.32 ± 0.06Mg xi 9.23 . . . . . . 0.26 ± 0.10 9.252 −430 ± 390 1600 ± 360 0.57 ± 0.06Ne ix Heγ 11.000 110 ± 220 . . . 0.33 ± 0.06Ne ix Heβ 11.547 . . . . . . .0.42 ± 0.07Ne x Lyα 12.134 −100 ± 200 470 ± 120 1.47 ± 0.15 12.206 640 ± 120 1200 ± 160 1.62 ± 0.08Fe xx 12.846 −500 ± 190 . . . 0.60 ± 0.15 12.888 −160 ± 210 710 ± 240 0.45 ± 0.06Ne ix r 13.447 −90 ± 180 . . . 1.59 ± 0.17 13.535† . . . . . . 2.31 ± 0.10Ne ix i 13.552 . . . . . . .0.83 ± 0.17Ne ix f 13.698 −30 ± 180 . . . 2.03 ± 0.20 13.756 130 ± 90 830 ± 110 1.81 ± 0.10O viii RRC 14.228 . . . . . . 1.25 ± 0.13 14.326† . . . . . . 2.52 ± 0.10O viii Lyδ 14.821 . . . . . . .0.37 ± 0.06Fe xvii 15.014 −60 ± 160 570 ± 100 1.67 ± 0.17 15.153† . . . . . . 3.10 ± 0.11O viii Lyγ 15.176 . . . . . . . . .Fe xvii 15.261 −210 ± 160 570 ± 100 0.65 ± 0.13O viii Lyβ 16.006 −90 ± 150 . . . 1.45 ± 0.15 16.075 150 ± 110 1100 ± 110 1.30 ± 0.07O vii RRC 16.78 . . . . . . 2.43 ± 0.24 16.806 . . . . . . 2.95 ± 0.10Fe xvii 17.051, 17.096 −180 ± 140 . . . 2.78 ± 0.28 17.142† . . . . . . 2.81 ± 0.12O vii Heδ 17.396 −520 ± 140 . . . 0.63 ± 0.07O vii Heγ 17.768 −380 ± 140 640 ± 80 0.79 ± 0.08 17.814 −360 ± 100 350 ± 220 0.52 ± 0.05N vii RRC 18.587 . . . . . . 1.14 ± 0.11O vii Heβ 18.627 −430 ± 130 . . . 1.32 ± 0.13 18.664 −540 ± 130 1570 ± 140 1.84 ± 0.09O viii Lyα 18.969 −200 ± 130 700 ± 80 5.89 ± 0.59 19.044 50 ± 30 690 ± 30 6.06 ± 0.14N vii Lyδ 19.361 −380 ± 120 . . . 0.12 ± 0.06N vii Lyγ 19.826 −350 ± 120 390 ± 80 0.44 ± 0.07N vii Lyβ 20.910 −370 ± 120 520 ± 70 0.81 ± 0.12 20.967 −320 ± 100 450 ± 150 0.83 ± 0.09O vii r 21.602 −260 ± 110 450 ± 70 4.50 ± 0.45 21.678 −90 ± 40 550 ± 50 4.95 ± 0.20O vii i 21.803 −380 ± 110 450 ± 70 1.75 ± 0.19 21.869 −230 ± 100 570 ± 120 2.02 ± 0.16O vii f 22.101 −430 ± 110 450 ± 70 9.59 ± 0.96 22.164 −290 ± 30 460 ± 40 10.08 ± 0.28N vi RRC 22.46 . . . . . . 2.06 ± 0.21 22.490 . . . . . . 1.46 ± 0.14N vi Heδ 23.277 −210 ± 100 470 ± 60 0.32 ± 0.14N vi Heγ 23.771 −490 ± 100 360 ± 60 0.77 ± 0.12 23.838 −290 ± 190 300 ± 200 0.36 ± 0.08N vii Lyα 24.781 −270 ± 100 650 ± 60 6.09 ± 0.74 24.876 10 ± 40 690 ± 40 6.18 ± 0.18C vi RRC 25.303 . . . . . . 2.83 ± 0.28 25.258 . . . . . . 2.69 ± 0.13C vi Lyδ 26.357 −440 ± 90 400 ± 60 0.57 ± 0.10 26.438 −220 ± 130 < 140 0.30 ± 0.06C vi Lyγ 26.990 −440 ± 90 320 ± 60 0.95 ± 0.11 27.073 −220 ± 60 280 ± 80 0.98 ± 0.08C vi Lyβ 28.466 −380 ± 80 450 ± 50 2.03 ± 0.20 28.541 −350 ± 60 230 ± 120 1.00 ± 0.10N vi r 28.787 −350 ± 80 410 ± 50 3.84 ± 0.38 28.882 −150 ± 40 330 ± 50 2.80 ± 0.16N vi i 29.083 −310 ± 80 410 ± 50 1.22 ± 0.15 28.933† . . . . . . 6.00 ± 0.25N vi f 29.534 −430 ± 80 410 ± 50 8.46 ± 0.85 29.622 −250 ± 20 420 ± 30 7.87 ± 0.22Fe xxiv O 30.746 . . . . . . . . . 30.827 . . . . . . 2.86 ± 0.18C v RRC 31.63 . . . . . . 4.30 ± 0.43 31.613 . . . . . . 7.55 ± 0.24S xiii O 32.239 . . . . . . . . . 32.331 . . . . . . 2.65 ± 0.17C v Heδ O 32.754 . . . . . . . . . 32.829 . . . . . . 0.68 ± 0.09C v Heγ 33.426 . . . 550 ± 50 1.50 ± 0.31 33.589 320 ± 140 820 ± 100 1.65 ± 0.18C vi Lyα 33.736 −360 ± 70 510 ± 40 12.29 ± 1.23 33.842 −200 ± 20 420 ± 20 9.48 ± 0.27Ar xi O 34.330 . . . . . . . . . 34.435 . . . . . . 0.84 ± 0.13C v Heβ 34.973 −550 ± 70 360 ± 40 1.28 ± 0.30 35.058 −410 ± 90 470 ± 80 1.18 ± 0.14S xii O 36.398 . . . . . . . . . 36.553 . . . . . . 1.11 ± 0.12Ca xii O 37.604 . . . . . . . . . 37.685 . . . . . . 1.84 ± 0.21

Notes. Measurements on the left-hand-side (Cols. 3−5) are from Kinkhabwala et al. (2002), while those on the right-hand-side (Cols. 6−9) arefrom the present work. voff is the velocity shift. σv is the velocity broadening. Fpho is the observed photon flux. Those line IDs labeled with a downtriangle (O) are inconclusive. Those labeled with dagger (†) correspond to multiple lines.

A122, page 14 of 16

J. Mao et al.: CIELO-RGS

Table A.2. List of lines in Procyon.

Line ID λ0 λR1 FR1 λR2 FR2 λL FL λR FR

Å Å ph m−2 s−1 Å ph m−2 s−1 Å ph m−2 s−1 Å ph m−2 s−1

Ne ix r 13.448 . . . . . . 13.454(9) 0.17(4) 13.450(13) 0.14(5) 13.561 ± 0.017† 0.41 ± 0.05Ne ix r 13.700 . . . . . . 13.701(6) 0.17(4) 13.690(15) 0.12(5)Fe xvii† 15.014 15.018(7) 0.20(5) 15.025(9) 0.26(4) 15.015(18) 0.21(6) 15.138 ± 0.016† 0.51 ± 0.04O viii 15.175 15.176(13) 0.12(3) 15.161(21) 0.08(3) . . . . . .

Fe xvii 15.265 15.281(12) 0.11(3) 15.258(12) 0.09(5) . . . . . .

Fe xvii 16.780 16.776(9) 0.16(3) 16.776(15) 0.12(4) 16.790(14) 0.13(5)Fe xvii 17.055 17.047(9) 0.23(4) 17.044(14) 0.19(5) 17.054(12) 0.22(9) 17.077 ± 0.005† 0.61 ± 0.05Fe xvii 17.100 . . . . . . 17.099(12) 0.25(5) 17.102(9) 0.29(9)O viii 17.380 17.402(12) 0.07(3) 17.402(28) 0.06(3) 17.396(17) 0.08(5)O viii 17.770 17.765(9) 0.10(3) 17.772(9) 0.11(4) 17.796(13) 0.14(5)O viii 18.628 18.624(4) 0.30(4) 18.637(6) 0.37(5) 18.629(5) 0.39(8) 18.638 ± 0.006 0.35 ± 0.04O viii 18.973 18.970(3) 1.61(8) 18.975(2) 1.78(11) 18.972(2) 1.83(15) 18.972 ± 0.006 2.38 ± 0.08N vii 20.910 20.198(27) 0.04(3) . . . . . . 20.905(22) 0.14(7)O vii 21.602 21.596(2) 2.36(11) . . . . . . 21.597(2) 3.01(25) 21.604 ± 0.002 4.08 ± 0.16O vii 21.800 21.797(4) 0.53(6) . . . . . . 21.792(5) 0.90(14) 21.807 ± 0.005 0.91 ± 0.09O vii 22.100 22.098(2) 2.20(11) . . . . . . 22.089(2) 2.57(23) 22.097 ± 0.004 3.54 ± 0.15N vii 24.781 . . . . . . 24.780(3) 0.88(7) 24.790(4) 0.80(14) 24.782 ± 0.006 1.17 ± 0.08N vi 24.900 . . . . . . 24.907(22) 0.08(4) 24.906(11) 0.18(9)C vi 26.990 27.001(11) 0.16(4) 26.994(12) 0.14(4) 26.979(19) 0.18(9) 27.003 ± 0.010 0.17 ± 0.03C vi 28.470 28.460(7) 0.30(5) 28.465(6) 0.39(6) 28.470(6) 0.49(12) 28.466 ± 0.005 0.64 ± 0.06N vi 28.790 28.785(4) 0.69(9) 28.775(6) 0.71(8) 28.785(5) 0.88(15) 28.789 ± 0.003 1.13 ± 0.08N vi 29.090 29.078(9) 0.24(5) 29.084(9) 0.29(7) 29.082(12) 0.29(10) 29.077 ± 0.009 0.41 ± 0.06N vi 29.530 29.524(6) 0.43(6) 29.520(8) 0.38(6) 29.546(11) 0.43(13) 29.523 ± 0.008 0.69 ± 0.07Ca xi 30.448 30.445(12) 0.20(5) 30.446(13) 0.22(5) 30.450(25) 0.18(11) 30.444 ± 0.009 0.35 ± 0.05Si xii 31.015 31.027(16) 0.18(5) 31.021(12) 0.15(5) 31.054(15) 0.22(10) 31.011 ± 0.014 0.37 ± 0.06S xivO 33.549 . . . . . . 33.490(26) 0.15(7) 33.510(18) 0.17(10)C vi 33.736 33.724(2) 3.49(17) 33.726(2) 4.15(30) 33.731(2) 4.02(32) 33.728 ± 0.001 6.38 ± 0.21C v 34.970 34.967(12) 0.19(6) 34.962(15) 0.17(7) 34.959(15) 0.27(17) 34.994 ± 0.021 0.25 ± 0.06Ca xi 35.212 35.198(30) 0.13(6) 35.193(12) 0.27(8) 35.188(18) 0.29(15) 35.202 ± 0.020 0.29 ± 0.07Ca xi 35.576 35.566(16) 0.15(7) 35.562(12) 0.26(7) 35.566(16) 0.23(14)S xiii 35.665 35.682(8) 0.37(8) 35.676(9) 0.38(8) 35.672(9) 0.53(16) 35.644 ± 0.010 0.79 ± 0.10S xii 36.398 36.374(12) 0.35(10) 36.372(15) 0.28(7) 36.399(15) 0.34(14) 36.454 ± 0.018† 0.82 ± 0.10S xii 36.563 36.544(19) 0.15(6) 36.561(19) 0.29(9) 36.547(15) 0.24(13)S xiiiO 36.72 . . . . . . . . . . . . . . . . . . 37.674 ± 0.065† 0.40 ± 0.12

Notes. Measurements on the left-hand-side (Cols. 3−8) are from Raassen et al. (2002), while those on the right-hand-side (Cols. 9 and 10) arefrom the present work. Those line IDs labeled with a down triangle (O) are inconclusive. Those labeled with dagger (†) correspond to multiplelines.

A122, page 15 of 16

A&A 625, A122 (2019)

Table A.3. List of lines for Her X-1 at different orbital phases.

Line ID λ0 State λobs σv Fpho λobs σv Fpho

Å Å km s−1 ph m−2 s−1 Å Å ph m−2 s−1

Ne x Lyα 12.134 a . . . . . . . . . . . . . . . . . .b . . . . . . 0.7 ± 0.3 12.19 ± 0.02 1300 ± 600 0.8 ± 0.2c . . . . . . 0.4 ± 0.3 . . . . . . . . .

Ne ix r 13.447 a . . . . . . . . . . . . . . . . . .b . . . . . . <0.5 . . . . . . . . .c . . . . . . 0.4 ± 0.3 . . . . . . . . .

Ne ix i 13.552 a . . . . . . . . . . . . . . . . . .b . . . . . . <0.5 13.50 ± 0.03 2000 ± 500 1.0 ± 0.2c . . . . . . 1.2 ± 0.4 . . . . . . . . .

Ne ix f 13.698 a . . . . . . . . . . . . . . . . . .b . . . . . . <0.16 . . . . . . . . .c . . . . . . <0.6 . . . . . . . . .

O viii Lyα 18.969 a 19.05 ± 0.05 P Cygni? .9 19.12 ± 0.02 1200 ± 500 6.0 ± 1.0b 18.96 ± 0.01 390 ± 200 1.2 ± 0.3 18.971 ± 0.008 370 ± 170 1.15 ± 0.13c 18.96 ± 0.02 <450 1.4 ± 0.3 18.967 ± 0.008 320 ± 240 1.4 ± 0.2

O vii r 21.602 a . . . 3200 ± 800 3.8 ± 0.8 . . . . . . . . .b 21.61 ± 0.03 <1600 1.0 ± 0.4 . . . . . . . . .c 21.62 ± 0.02 . . . 0.8 ± 0.4 . . . . . . . . .

O vii i 21.803 a . . . 3200 ± 800 16 ± 3 21.77 ± 0.03 2200 ± 400 11.6 ± 1.5b 21.79 ± 0.01 <400 3.1 ± 0.5 21.785 ± 0.005 220 ± 220 3.0 ± 0.3c 21.82 ± 0.02 320 ± 160 7.0 ± 1.0 21.801 ± 0.004 250 ± 150 5.9 ± 0.4

O vii f 22.101 a . . . . . . . . . . . . . . . . . .b . . . . . . <0.16 . . . . . . . . .c . . . . . . <0.16 . . . . . . . . .

N vii Lyα 24.781 a 24.85 ± 0.17 3200 (fixed) 8.4 ± 0.3 24.80 ± 0.08 4200 ± 1600 12.9 ± 2.4b 24.77 ± 0.02 <590 1.5 ± 0.3 24.785 ± 0.005 100 ± 160 1.7 ± 0.2c 24.77 ± 0.02 <590 2.3 ± 0.6 24.777 ± 0.008 600 ± 140 3.2 ± 0.3

N vi r 28.787 a . . . . . . . . . . . . . . . . . .b 28.77 ± 0.01 <260 0.7 ± 0.2 28.783 ± 0.010 <150 0.8 ± 0.2c 28.75 ± 0.02 <260 0.9 ± 0.5 . . . . . . . . .

N vi i 29.083 a . . . . . . .3 . . . . . . . . .b 29.07 ± 0.01 270 ± 100 2.9 ± 0.4 29.083 ± 0.005 <110 4.0 ± 0.3c 29.10 ± 0.02 <260 4.0 ± 0.7 29.075 ± 0.004 220 ± 120 6.0 ± 0.4

N vi f 29.534 a . . . . . . . . . . . . . . . . . .b . . . . . . <0.09 . . . . . . . . .c . . . . . . <0.16 . . . . . . . . .

C vi Lyα 33.736 a . . . . . . . . . . . . . . . . . .b 33.72 ± 0.01 <380 0.6 ± 0.2 33.73 ± 0.02 <2100 0.8 ± 0.2c 33.72 ± 0.02 <320 1.0 ± 0.4 33.73 ± 0.02 340 ± 190 1.4 ± 0.3

Notes. The main-on (orbital phase 0.18−0.26), low (0.47−0.56), and short-on (0.52−0.60) states are noted as a (0134120101), b (0111060101),and c (0111061201). Measurements on the left-hand-side (Cols. 4−6) are from Jimenez-Garate et al. (2002) and errors are 90% confidence limits.Those on the right-hand-side (Cols. 7−9) are from the present work and errors are 68% confidence limits.

A122, page 16 of 16